The power dissipation of integrated circuit chips, and the modules containing the chips, continues to increase in order to achieve increases in processor performance. This trend poses a cooling challenge at both the module and system level. Increased air flow rates are needed to effectively cool high power modules and to limit the temperature of the air that is exhausted into the computer center.
In many large server applications, processors along with their associated electronics (e.g., memory, disk drives, power, etc.) are packaged in removable drawer configurations stacked within a rack or frame. In other cases, the electronics may be in fixed locations within the rack or frame. Typically, the components are cooled by air moving in parallel air flow paths, usually front-to-back, impelled by one or more air moving devices (e.g., fans or blowers). In some cases it may be possible to handle increased power dissipation within a single drawer by providing greater air flow, through the use of a more powerful air moving device or by increasing the rotational speed (i.e., RPMs) of an existing air moving device. However, this approach is becoming problematic at the rack level in the context of a computer installation (e.g., data center).
The sensible heat load carried by the air exiting the rack is stressing the ability of the room air conditioning to effectively handle the load. This is especially true for large installations with “server farms” or large banks of computer racks close together. In such installations not only will the room air conditioning be challenged, but the situation may also result in re-circulation problems with some fraction of the “hot” air exiting one rack unit being drawn into the air inlet of the same rack or a nearby rack. This re-circulating flow is often extremely complex in nature, and can lead to significantly higher rack inlet temperatures than expected. This increase in cooling air temperature may result in components exceeding their allowable operating temperature or in a reduction in long term reliability of the components.
Addressing thermal imbalances within a data center is often an expensive and time consuming operation. Therefore, there remains a need in the art for an apparatus and method which facilitate thermal and energy-based design and optimization of a data center.